"""
Demo is based on the [Kernel PCA] - (https://scikit-learn.org/stable/auto_examples/decomposition/plot_kernel_pca.html#sphx-glr-auto-examples-decomposition-plot-kernel-pca-py
"""

from sklearn.datasets import make_circles
from sklearn.model_selection import train_test_split
from sklearn.decomposition import PCA, KernelPCA
import matplotlib
matplotlib.use('agg')
import matplotlib.pyplot as plt
import gradio as gr

def fit_plot(kernel, gamma, alpha, degree, coef0):
    X, y = make_circles(n_samples=1_000, factor=0.3, noise=0.05, random_state=0)
    X_train, X_test, y_train, y_test = train_test_split(X, y, stratify=y, random_state=0)
    pca = PCA(n_components=2)
    if kernel=="linear":
        kernel_pca = KernelPCA(n_components=2, kernel=kernel, fit_inverse_transform=True, alpha=alpha)
    elif kernel=="poly":
        kernel_pca = KernelPCA(n_components=2, kernel=kernel, gamma=gamma, degree=degree, coef0=coef0, fit_inverse_transform=True, alpha=alpha)
    elif kernel=="rbf":
        kernel_pca = KernelPCA(n_components=2, kernel=kernel, gamma=gamma, fit_inverse_transform=True, alpha=alpha)
    elif kernel=="cosine": 
        kernel_pca = KernelPCA(n_components=2, kernel=kernel, fit_inverse_transform=True, alpha=alpha)
    X_test_pca = pca.fit(X_train).transform(X_test)
    X_test_kernel_pca = kernel_pca.fit(X_train).transform(X_test)
    fig, (orig_data_ax, pca_proj_ax, kernel_pca_proj_ax) = plt.subplots(ncols=3, figsize=(14, 4))
    orig_data_ax.scatter(X_test[:, 0], X_test[:, 1], c=y_test)
    orig_data_ax.set_ylabel("Feature #1")
    orig_data_ax.set_xlabel("Feature #0")
    orig_data_ax.set_title("Testing data")
    pca_proj_ax.scatter(X_test_pca[:, 0], X_test_pca[:, 1], c=y_test)
    pca_proj_ax.set_ylabel("Principal component #1")
    pca_proj_ax.set_xlabel("Principal component #0")
    pca_proj_ax.set_title("Projection of testing data\n using PCA")
    kernel_pca_proj_ax.scatter(X_test_kernel_pca[:, 0], X_test_kernel_pca[:, 1], c=y_test)
    kernel_pca_proj_ax.set_ylabel("Principal component #1")
    kernel_pca_proj_ax.set_xlabel("Principal component #0")
    _ = kernel_pca_proj_ax.set_title("Projection of testing data\n using KernelPCA")
    return fig

with gr.Blocks() as demo:
    gr.Markdown("## PCA vs Kernel PCA")
    #state = gr.State([])
    with gr.Row(variant='panel').style(equal_height=True):
        p1 = gr.Dropdown(choices=["linear", "poly", "rbf", "cosine"], label="Kernel", value="rbf", interactive=True)
    with gr.Row(variant='panel').style(equal_height=True):
        p2 = gr.Slider(0, 10, label="Kernel coefficient (for rbf, poly and sigmoid kernels)", value=None, step=1e-3, interactive=True)
        p3 = gr.Slider(0, 1, label="Alpha of ridge regression (for non-precomputed kernels)", value=1, step=1e-3, interactive=True)
    with gr.Row(variant='panel').style(equal_height=True):
        p4 = gr.Slider(0, 10, label="Degree (for poly kernel)", value=3, step=1, interactive=True)
        p5 = gr.Slider(0, 10, label="Independent term (for poly and sigmoid kernels)", value=1, step=1e-1, interactive=True)
    btn = gr.Button(value="Submit")
    out = gr.Plot(label="Projecting data with PCA and Kernel PCA")
    btn.click(fit_plot, inputs=[p1,p2,p3,p4,p5], outputs=out)

demo.launch()